10 ms·
"Log drain" appears to be a Heroku term of art for something that can receive shipped log data from their infrastructure, so an "HTTP log drain front-end" would
by throwanem 10y ago
"Log drain" appears to be a Heroku term of art for something that can receive shipped log data from their infrastructure, so an "HTTP log drain front-end" would be a client; if that or something like it is in play here, it would explain why they don't host the aggregator in the same place as whatever's generating the log entries it ingests.
40 requests/second does sound somewhat low, but that may just be the average rate at which whatever they're aggregating generates log entries, rather than a limit enforced by resource exhaustion.